In the highly competitive content provider industry, content providers have much interest in identifying which contents users are interested in and trends of users' interest in content. Conventionally, ratings groups have users complete detailed surveys to gather information about what content the users are interested in. The ratings groups then calculate a rating for each of the contents based on the number of users is interested in a particular content. Also, rating groups monitor a number of times a keyword has been mentioned in contents in order to identify a trend in the contents. Ratings surveys and monitoring services, however, may not accurately reflect what contents users are interested in over time at least due to a limited sample size, limited time period, limited parameters in consideration, and errors in users completing the surveys.